Maximum entropy estimation of probability distributions with Gaussian conditions

Abstract

We describe a method to computationally estimate the probability density function of a univariate random variable by applying the maximum entropy principle with some local conditions given by Gaussian functions. The estimation errors and optimal values of parameters are determined. Experimental results are presented. The method estimates the distribution well if a large enough selection is used, typically at least 1 000 values. Compared to the classical approach of entropy maximisation, local conditions allow improving estimation locally. The method is well suited for a heuristic optimisation approach.